Introduction
Down syndrome (DS, OMIM #190685) is the most common genetic cause of intellectual developmental disorder worldwide, yet actionable early-life interventions capable of modifying long-term neurodevelopmental trajectories remain extremely limited. Individuals with DS also experience congenital malformations and a range of specific comorbidities, for which early prevention and treatment can significantly improve quality of life and life expectancy. Among these, obstructive sleep apnea (OSA) is highly prevalent in children with DS, affecting up to 90% of this population [1, 2].
This high susceptibility is driven by craniofacial dysmorphism (midfacial hypoplasia, brachycephaly, micrognathia), an abnormally small upper airway, glossoptosis with relative macroglossia, pharyngeal hypotonia, pharyngolaryngomalacia, and adenotonsillar hypertrophy, exacerbated by a genetic susceptibility to infections [3].
OSA-related sleep disturbances, including intermittent hypoxemia and sleep fragmentation, adversely affect neurocognitive development and daytime behavior. This is particularly concerning in children with DS who already exhibit baseline neurodevelopmental impairment [4, 5]. Persistent OSA may exacerbate deficits in executive function, attention, and memory, while contributing to behavioral disturbances such as hyperactivity, aggression, and internalizing symptoms [5, 6].
In the prospective interventional Respire−21 study, we demonstrated that early recognition and treatment of OSA were associated with improved neurocognitive and behavioral outcomes at 36 months of age in infants with DS [7]. The study enrolled an unselected cohort of infants with DS aged 6 months (n = 39) who underwent systematic polysomnography (PSG) every 6 months (screened group). The participants diagnosed with OSA received personalized OSA treatment.
At 36 months of age, the participants in the screened group underwent a final PSG and neurocognitive evaluation; their development and adaptive behavior were then compared with those of a control group of children with DS receiving standard care (n = 40; standard care group). Standard care included PSG only when a sleep disorder was clinically suspected; otherwise, PSG screening was recommended for all children with DS between the ages of 3 and 4 years [8].
The prevalence of OSA in our cohort was high (≥ 90%), consistent with the findings of epidemiologic studies in children with DS [2, 9]. At 36 months of age, the screened group showed a superior primary neurocognitive outcome (Griffiths III Global Quotient of Development) compared with the standard care group. This finding was further supported by improvements across Griffiths III domain subscores; Vineland Adaptive Behavior Scales, Second Edition (VABS-II) scores; PSG parameters; and sleep architecture of the screened group [7].
However, it remains unclear whether early treatment of OSA is associated with sustained benefits on the neurocognitive development of children with DS, particularly with respect to language skills. Studies in typically developing children have shown that daytime naps and nighttime sleep support long-term retention of language structures and grammar and facilitate the acquisition and integration of new vocabulary [1, 10–12]. However, the relationship between sleep disturbances and language development remains underexplored in children with DS [4, 5, 13].
Therefore, this follow-up study aimed to determine the potential long-term benefits of early detection and treatment of OSA on the cognitive outcomes of children with DS at 60 months of age compared with the cognitive outcomes of the children in the standard care group and to evaluate the impact of early OSA intervention on language and communication development in children with DS.
Methods
Study design
The design and results of the initial study have been reported elsewhere [7]. All children diagnosed with OSA and treated before 36 months of age (screened group) or from 36 months onward (standard care group) in the initial study were invited to participate in this 2-year follow-up study. This study included a single follow-up at 60 months of age.
Information regarding socioeconomic factors, childcare, school attendance, bilingualism, and receipt of supportive therapies (e.g., speech therapy, physiotherapy, psychomotor therapy, and palatal plate therapy) were collected from the participants, and clinical examination, hearing and vision screening, neuropsychological assessment, and PSG were performed.
Ethical statement
The study was conducted in accordance with French regulatory requirements and was approved by the relevant ethics committee (ClinicalTrials.gov Identifier NCT03210675).
Participants
Children from the screened group and standard care group were eligible for the follow-up study if they met the following criteria:Screened group: Children with an apnea–hypopnea index (AHI) of ≥ 1 event/h on at least one of the PSGs before the 36-month PSG and who received OSA treatment, i.e., upper airway surgery, pharmacological therapy with topical corticosteroids and/or oral leukotriene receptor antagonists administered for at least 1 month (± 2-day margin), and/or continuous positive airway pressure (CPAP), before the age of 36 months.Standard care group: Children with an AHI of ≥ 1 event/h at the 36-month PSG and who underwent OSA treatment.
Endpoints
All outcomes in this follow-up study were predefined as secondary endpoints. Neuropsychological assessments were performed at the Institut Jérôme Lejeune, Paris, by trained neuropsychologists blinded to the PSG findings. Griffiths III was administered at 36 and 60 months of age [14].
Executive functioning and behavioral outcomes of the children were evaluated using parent-completed Behavior Rating Inventory of Executive Function–Preschool (BRIEF-P) and Child Behavior Checklist-Preschool (CBCL-P). Vineland Adaptive Behavior Scales, Second Edition (VABS-II) was administered through a semi-structured interview conducted with the parents by the neuropsychologists [15–17]. Language outcomes at 60 months of age were evaluated using the Griffiths III Language and Communication subscale; the raw scores of the Picture Naming and Receptive Vocabulary tests of the Wechsler Preschool and Primary Scale of Intelligence, Fourth Edition (WPPSI-IV), a verbal fluency task (assessed via the number of correct words generated within 1 min); and the language hetero-questionnaire provided in the CBCL-P (mean sentence length and vocabulary score; Table S1) [14, 16, 18].
The PSG outcomes included variables associated with sleep duration, sleep-stage distribution, respiratory event frequency (apnea and hypopnea), and gas exchange parameters.
Polysomnography
PSG at 60 months of age was performed in room air at the sleep unit of Necker Enfants Malades Hospital using the CID 102* system (Cidelec, Saint Gemmes sur Loire, France) or the Alice 6 system (Philips Respironics, Lyon, France). The PSG recordings were independently scored by three pediatric sleep experts according to the American Academy of Sleep Medicine guidelines [19]. OSA severity was classified according to the AHI as follows: no OSA (AHI < 1 event/h), mild OSA (AHI 1–5 events/h), moderate OSA (AHI > 5–10 events/h), and severe OSA (AHI > 10 events/h).
Statistical analysis
Descriptive statistics were used to summarize participant characteristics at 36 months of age and neuropsychological outcomes at 36 and 60 months of age in the NEURO-60M population, defined as all children enrolled in the follow-up study who had been diagnosed with OSA before 36 months (screened group) or at 36 months (standard care group) and underwent neuropsychological evaluation at both timepoints. The same analyses were performed in the NEURO-60M per-protocol (PP) population, which included children from the NEURO-60M population who received OSA treatment and underwent no major protocol deviations (Fig. 1).
For continuous variables, the number of observed and missing values; median, first and third quartiles; and minimum and maximum values were reported. For categorical variables, the number of observed and missing values and the frequency and percentage of participants in each category were summarized. Between-group differences were estimated using the Hodges–Lehman estimator, and corresponding p values were calculated using the Wilcoxon rank-sum test.
Between-group comparisons for secondary endpoints were not adjusted for multiplicity. Descriptive analyses of PSG outcomes were performed for the PSG-60M population, defined as all children diagnosed with OSA before 36 months (screened group) or at 36 months of age (standard care group) who had available PSG data at 60 months of age. All statistical analyses were conducted using SAS® software (version 9.4).Fig.
1Study flowchart. IS-60M, all children who participated in the 36–60 months follow-up study. NEURO-60M, all children in the IS-60M set who had a diagnosis of OSA either before 36 months (screened group) or at 36 months (standard care group) and who underwent an NP assessment at 36 and 60 months.
NEURO-60M PP, all children in the NEURO-60M set who were treated for OSA and had no other major protocol deviations. PSG-60M, all children in the IS-60M set who had been diagnosed with OSA before 36 months (screened group) or at 36 months (standard care group) and who had available 60-month PSG results. Participants whose 36-month NP assessment was performed before 34 months or after 39 months of age or whose 60-month NP assessment was performed before 57 months or after 65 months of age were not included in the NEURO-60M PP population.
60M, 60 months; IS, included set; NP, neuropsychological; OSA, obstructive sleep apnea; PP, per-protocol; PSG, polysomnography; DS, Down syndrome
Results
Participants
The initial study included 79 infants and children with DS (screened group: n = 39; standard care group: n = 40). Of these, 29 and 18 participants from the screened group and standard care group, respectively, completed the 60-month follow-up visit, comprising the included set (IS) — 60M population (Fig. 1).
From the NEURO-60M PP population, five children in each group were excluded due to major protocol deviations (screened group: n = 24; standard care group: n = 13). From the PSG-60M population, one child in each group was excluded because of unavailable PSG data or central apnea (screened group: n = 28; standard care group: n = 17; Fig. 1).
Characteristics at 36 months of age, including anthropological measures, medical history, and use of rehabilitation therapies, particularly speech and palatal plate therapies, were comparable between the groups. Educational setting and bilingual exposure were also similar (Tables 1 and S2). Table 1Clinical characteristics and rehabilitation therapies of participants at 36 months of age (NEURO-60M PP population)ParameterScreened groupN = 24Standard care groupN = 13Age (months) Median36.035.0 Q1, Q335.0, 36.534.0, 36.0 Male, n (%)10 (41.7)4 (30.8)Head circumference (cm) Median47.446.5 Q1, Q346.8, 48.346.0, 47.0 BMI (z-score) Median0.7 − 0.2 Q1, Q3 − 0.3, 1.4 − 0.6, 0.7Associated disordersa, n (%) Cardiopathy18 (75.0)10 (76.9) Ventricular septal defect4 (16.7)2 (15.4) Atrial septal defect11 (45.8)4 (30.8) Ventricular atrial communication1 (4.2)1 (7.7) Fallot’s tetralogy1 (4.2)0 Duodenal atresia1 (4.2)0 Congenital hypothyroidism1 (4.2)0 Treated hypothyroidism1 (4.2)4 (30.8) Congenital cataract01 (7.7) Treated GERD5 (20.8)1 (7.7)Motor physiotherapy Yes, n (%)6 (25.0)5 (38.5) Speech therapy Yes, n (%)17 (70.8)10 (76.9) Psychomotor therapy Yes, n (%)13 (54.2)8 (61.5) Palatal plate therapybYes, n (%)2 (8.7)1 (7.7)BMI body mass index, ENT ear, nose, and throat, GERD gastro-esophageal reflux disease, Q1 first quartile, Q3 third quartileaOnly congenital diseases or ENT disorders (other than OSA), neurological and/or cardio-pulmonary diseases occurring during the first 36 months of life are shownbOne patient in the Screened group had missing data for palatal plate therapy
Neurocognitive development at 36 and 60 months of age (NEURO-60M PP population)
At 36 months of age, the median (Q1 and Q3) Griffiths III Global Quotient of Development (GQD) was 54.9 (52.0 and 60.0) in the screened group and 54.3 (50.0 and 56.8) in the standard care group (Fig. 2; Table S3). The estimated between-group difference was 2.5 points (95% CI − 1.7 to 7.3) in favor of the screened group but did not reach statistical significance (p = 0.323).
Similar findings were observed across the Griffiths III subscales (Fig. 2; Table S3).Fig. 2Box plot of Griffiths III Global Quotient of Development and selected subscale scores at 36 and 60 months of age (Neuro-60M PP population).
In each box plot, the box represents the interquartile range (IQR, 25th to 75th percentiles), the horizontal line indicates the median, and the cross (x) indicates the mean. Whiskers extend to the minimum and maximum values within 1.5 × IQR from the box boundaries. Griffiths III, Griffiths Scales of Child Development, Third Edition; QD, quotient of development
At 60 months of age, the median (Q1 and Q3) Griffiths III GQD was 47.5 (42.6 and 52.0) in the screened group and 42.9 (41.7 and 48.3) in the standard care group, corresponding to an estimated between-group difference of 3.9 points (95% CI − 0.9 to 10.0; p = 0.130; Fig. 2; Table S3). Among the Griffiths III domains, Foundations of Learning and Personal–Social–Emotional scores were significantly higher in the screened group than in the standard care group, with estimated between-group differences of 7.5 points (95% CI 1.6 to 13.8; p = 0.014) and 10.0 points (95% CI 2.0 to 18.0; p = 0.019), respectively (Fig.
2, Table S3).
At 36 months of age, the VABS-II global composite score and all subscale scores, except Motor Skills, were significantly higher in the screened group (Fig. 3; Table S3) than in the standard care group, with a VABS-II global composite score between-group difference of 12.0 points (95% CI 3.0 to 21.0; p = 0.008). A comparable between-group difference was maintained at 60 months of age (12.0 points; 95% CI 4.0 to 20.0; p = 0.007).
The VABS-II subscales were also significantly better in the screened group than in the standard care group, except for Communication and Motor skills (Fig. 3; Table S3).Fig. 3Box plot of VABS-II global composite score and selected subscales at 36 and 60 months of age (Neuro-60M PP population).
In each box plot, the box represents the interquartile range (IQR, 25th to 75th percentiles), the horizontal line indicates the median, and the cross (x) indicates the mean. Whiskers extend to the minimum and maximum values within 1.5 × IQR from the box boundaries. VABS-II, Vineland Adaptive Behavior Scales, Second Edition
No statistically significant between-group differences were observed in the BRIEF-P scores at either 36 or 60 months of age. A similar pattern was observed for the CBCL-P scores. However, analysis of syndromic scale scores at 60 months of age indicated more favorable outcomes in the screened group for the Somatic Complaints, Withdrawn, and Stress Problems domains compared with the standard care group with 23 children (95.8%) within the normal range (< 65) vs.
9 (69.2%), 22 (91.7%) vs. 7 (53.8%), and 23 (95.8%) vs. 9 (69.2%) respectively.
Notably, the majority of children had BRIEF-P and CBCL-P T-scores < 65. An exception was the BRIEF-P Emerging Metacognition Index (assessing attentional processes and problem-solving), for which elevated scores (≥ 65) were observed in 13 children (61.9%) in the screened group and 7 children (70.0%) in the standard care group at 60 months of age.
Language skills at 60 months of age (NEURO-60M PP population)
At 60 months of age, no significant between-group differences were observed for either the Griffiths III Language and Communication Quotient of Development (QD) or the VABS-II Communication subscale, with estimated between-group differences of 0.6 points (95% CI − 7.0 to 8.0; p = 0.874) and 3.5 points (95% CI − 7.0 to 14.0; p = 0.576; Table S3), respectively. Findings from the CBCL-P mean length of sentence, WPPSI-IV Receptive Vocabulary raw score, and verbal semantic fluency task were generally consistent with the absence of meaningful between-group differences. Alternatively, median CBCL-P vocabulary scores and median WPPSI-IV Picture Naming raw scores were lower in the screened group than in the standard care group (Table 2).
Table 2Language outcomes at 60 months of age (NEURO-60M PP population)ParameterScreened groupN = 24Standard care groupN = 13CBCL-P — mean length of sentence n (%)23 (95.8)13 (100) Median2.802.60 Q1, Q32.00, 3.001.00, 4.00CBCL-P — vocabulary score n (%)23 (95.8)13 (100) Median67.0112.0 Q1, Q327.0, 164.064.0, 151.0WPPSI-IV — picture naming raw score n (%)21 (87.5)10 (76.9) Median0.04.5 Q1, Q30.0, 5.04.0, 9.0WPPSI-IV — receptive vocabulary raw score n (%)22 (91.7)9 (69.2) Median8.58.0 Q1, Q36.0, 10.06.0, 10.0Verbal semantic fluency task — number of words in 1 min n (%)13 (54.2)9 (69.2) Median1.00.0 Q1, Q30.0, 2.00.0, 2.0Verbal semantic fluency task—number of correct words (among children with at least 1 word in 1 min) N74 Median2.02.5 Q1, Q31.0, 7.01.5, 3.5CBCL-P Child Behavior Checklist-Preschool, Q1 first quartile, Q3 third quartile, WPPSI-IV Wechsler Preschool and Primary Scale of Intelligence, Fourth Edition
PSG data at 36 and 60 months of age (PSG-60M population)
Seven children in the screened group did not undergo PSG at 36 months of age because of the COVID-19 epidemic, during which time an additional PSG would have been an added burden on children and their families (Table 3). At 36 months of age, the AHI was significantly higher in the standard care group than in the screened group, consistent with the findings of the primary analysis.7 Median (Q1 and Q3) AHI was 2.0 (1.0 and 4.0) events/h in the screened group compared with 11.0 (6.0 and 14.0) events/h in the standard care group (p < 0.001; Table 3). Table 3Polysomnographic outcomes at 36 and 60 months of age (PSG-60M population)Parameter36 Months60 MonthsScreened groupN = 28Standard care groupN = 17Screened groupN = 28Standard care groupN = 17With PSG, n (%)21 (75)17 (100)28 (100)17 (100)Age (months) Median37.037.060.060.0 Q1, Q337.0, 38.036.0, 37.060.0, 60.060.0, 60.0AHI (events/h) Median2.011.01.02.0 Q1, Q31.0, 4.06.0, 14.01.0, 2.51.0, 3.0 <1 events/h, n (%)1 (4.8)06 (21.4)2 (11.8) ≥1 to ≤5 events/h, n (%)17 (81.0)2 (11.8)20 (71.4)15 (88.2) >5 to ≤10 events/h, n (%)1 (4.8)6 (35.3)1 (3.6)0 >10 events/h, n (%)2 (9.5)9 (52.9)1 (3.6)0OAI (events/h) Median0.404.000.100.30 Q1, Q30.17, 0.822.25, 7.000.00, 0.400.00, 0.80 <1.5 events/h, n (%)18 (85.7)2 (11.8)26 (92.9)15 (88.2) ≥1.5 to ≤5 events/h, n (%)2 (9.5)8 (47.1)1 (3.6)2 (11.8) >5 to ≤10 events/h, n (%)05 (29.4)00 >10 events/h, n (%)1 (4.8)2 (11.8)1 (3.6)0CAI (events/h) Median0.250.270.100.10 Q1, Q30.10, 0.600.00, 0.930.00, 0.300.00, 0.30Total sleep time (TST) (min) Missing data0013 Median506.0448.0540.5515.3 Q1, Q3489.0, 570.0398.0, 520.0503.0, 599.5466.0, 614.0Sleep efficiency (%) Missing data0013 Median88.087.086.285.4 Q1, Q385.0, 91.079.0, 89.079.1, 91.475.6, 92.6WASO (min) Missing data0013 Median67.081.082.5100.5 Q1, Q342.0, 91.059.0, 133.051.5, 109.346.0, 171.7Sleep stages (%TST) Missing data0013 Sleep stage N1, median (Q1, Q3)6.00 (4.40, 7.80)11.00 (6.00, 14.00)5.70 (4.60, 7.30)6.25 (4.80, 8.50) Sleep stage N2, median (Q1, Q3)47.70 (43.00, 52.40)43.70 (35.80, 51.00)45.70 (41.80, 50.40)50.85 (47.80, 54.80) Sleep stage N3, median (Q1, Q3)25.40 (20.40, 28.80)28.00 (26.70, 33.60)25.90 (23.40, 30.80)24.90 (22.40, 29.20) Sleep stage REM, median (Q1, Q3)19.80 (16.70, 22.90)16.00 (6.70, 20.10)20.10 (16.40, 25.20)17.30 (13.10, 19.70)Mean SpO2 (%) Median96.096.096.097.0 Q1, Q396.0, 97.095.0, 97.096.0, 97.097.0, 98.0Minimal SpO2 (%) Median89.088.088.590.0 Q1, Q386.0, 90.084.0, 90.086.0, 90.588.0, 92.0Time with SpO₂ < 90% (%)<2%, n (%)20 (92.5)15 (88.2)28 (100)17 (100)≥2%, n (%)1 (4.8)2 (11.8)00Oxygen desaturation index ≥3% (events/h) Median5.012.05.44.6 Q1, Q33.0, 9.04.0, 19.02.9, 9.72.8, 5.7Mean PtcCO2 (mmHg) Missing data13995 Median42.541.040.042.0 Q1, Q339.5, 44.035.5, 43.535.0, 42.040.0, 44.0Maximal PtcCO2 (mmHg) Missing data13895 Median44.546.044.046.0 Q1, Q342.5, 47.543.0, 47.039.0, 45.043.5, 47.0Time with PtcCO2 > 50 mmHg (%) Missing data13795 <2%8 (100)10 (100)19 (100)12 (100)Polysomnographic results including apnea–hypopnea index, sleep parameters, and gas exchange parameters in the screened group and the standard care group at 36 and 60 months of age.
Categorization of OSA based on AHI: <1 events/h: no OSA; ≥1 to ≤5 events/h: mild OSA; >5 to 10 events/h: moderate OSA; >10 events/h: severe OSA. AHI apnea-hypopnea index; CAI central apnea index, OAI obstructive apnea index, PSG polysomnography, PtcCO2 transcutaneous carbon dioxide pressure, Q1 first quartile, Q3 third quartile, REM rapid eye movement, SpO2 pulse oximetry, WASO wake after sleep onset
At 60 months of age, the AHI remained stable in the screened group and improved in the standard care group. Most children had no OSA (21.4% vs 11.8% in the screened group and standard care group, respectively) or mild OSA (71.4% vs 88.2% in the screened group and standard care group, respectively; Table 3).
At 36 months of age, children in the PSG-60M population from the screened group exhibited a longer total sleep time than those in the standard care group (506 min vs 448 min; p = 0.020). They also spent a lower proportion of sleep time in stage N1 (p = 0.007) and a greater proportion in rapid eye movement (REM) sleep (p = 0.025) (Table 3). At 60 months of age, between-group differences were attenuated.
Total sleep time increased in both the groups (541 min vs 515 min in the screened group and standard care group, respectively; p = 0.630), and the distribution of sleep stages was generally comparable, except for a higher proportion of REM sleep in the screened group (20% vs 17%; p = 0.031).
OSA treatment during the follow-up study
In the screened group, only three of 29 participants (10%) underwent adenotonsillectomy for OSA during the follow-up period, with one participant additionally requiring turbinoplasty and glossectomy. In the standard care group, 16 of 18 participants (89%) underwent upper airway surgery for OSA, all of whom had adenoidectomy. Nine additionally underwent tonsillectomy only and seven underwent at least two additional procedures.
Regarding non-surgical interventions, CPAP was used during the follow-up period by one participant in the screened group and three participants in the standard care group. Furthermore, pharmacological therapy with potential activity on OSA (topical corticosteroids and/or oral leukotriene receptor antagonists) was received by seven participants in the screened group and six in the standard care group, although these treatments may have been initiated for indications other than OSA (such as allergic rhinitis or asthma).
Discussion
This follow-up study demonstrated that early screening and treatment of OSA in infants with DS was associated with long-term neurocognitive benefits at 60 months of age, as shown by the Griffiths III Foundations of Learning and Personal–Social–Emotional domains, and the VABS-II subscale scores for Daily Living Skills, and Socialization. However, these benefits did not extend to language outcomes assessed using multiple complementary measures.
Previous studies in both typically developing children and older children with DS have reported associations between sleep-disordered breathing and/or OSA and impairments across cognitive domains, including memory and language development [13, 20–22]. Differences between those findings and our results may reflect variations in study design, age range, and population characteristics, as few studies have specifically focused on young children with DS. In a cross-sectional study of 30 children with DS aged 6–18 years, the proportion of REM sleep was significantly correlated with performance on the WPPSI-Revised Vocabulary measure and the VABS-II Communication-Receptive, Communication-Expressive subscales [22].
Another study categorized children with DS aged 26–64 months as “poor sleepers” and “good sleepers” based on actigraphy-derived sleep efficiency below or above 80%, respectively [13]. Typically developing children and children with DS classified as good sleepers consistently achieve better scores on the MacArthur Communicative Development Inventory than children with DS classified as poor sleepers [13]. Alternatively, a cross-sectional study comparing children with DS aged 6–17 years with no OSA (n = 38), untreated OSA (n = 28), or treated OSA (n = 34) found no significant differences in vocabulary outcomes, including performance on the Peabody Picture Vocabulary Test, fifth edition, and the Expressive Vocabulary Test, third edition [5].
The lack of benefits in language and communication outcomes in our study may have several explanations. Expressive language domains, particularly phonology and syntax, represent a well-recognized area of difficulty in children with DS, who typically exhibit greater deficits and delays in these domains than in other aspects of cognition [23–25]. Therefore, assessment at 60 months of age may be too early to capture the potential downstream effects of OSA treatment or sleep architecture on expressive language development.
Although as observed in the screened group, early screening and treatment of OSA is associated with a moderate but statistically significant advantage in the proportion of REM sleep at both 36 and 60 months of age, this difference was not accompanied by measurable improvements in language or communication outcomes relative to the standard care group. Additional characteristics of language development in children with DS should also be considered. During early lexical development, children with DS rely more heavily on gestural than spoken communication, with gesture use typically declining with increasing oral language production, raising the possibility that the transition from compensatory gestural communication to spoken language may be influenced by sleep quality [26].
However, the cognitive and neural mechanisms underlying language modality preference in infants with DS remain poorly understood.
The persistence of favorable neurocognitive outcomes in the screened group at 60 months of age further supports the value of systematic early screening and treatment of OSA in children with DS. Although the between-group difference in Griffiths III GQD at 60 months of age did not reach statistical significance, its magnitude remained comparable to that observed at 36 months of age in the original analysis (36-month original analysis: 4.1 points; p = 0.009; 60-month follow-up study: 3.9 points; p = 0.130) [7]. The lack of statistical significance at 60 months of age was anticipated, given the reduced sample size of the extension cohort, which included only children who received treatment for OSA.
A significant between-group difference persisted at 60 months of age in the Griffiths III Personal–Social–Emotional domain, with an estimated difference of 10.0 points (p = 0.019), while the difference in the Foundations of Learning domain—which previously showed a strong trend at 36 months—reached statistical significance, with an estimated difference of 7.5 points (p = 0.014). The Personal–Social–Emotional domain reflects the child’s ability to recognize emotions and establish social relationships [27], while the Foundations of Learning domain, particularly impaired in children with DS, captures attentional, executive, reasoning, and memory functions. Together, all of these are essential to neurocognitive development during early childhood.
These findings were further supported by the VABS-II results, in which the largest between-group differences at 60 months of age were observed in the Daily Living Skills (16.0 points;p = 0.004) and Socialization (17.0 points; p =0.005) domains. Interestingly, Soltani et al. identified significantly better outcomes in the BRIEF2 Emotional Regulation Index and the CBCL Internalizing subdomain among children with DS who had no OSA or received OSA treatment compared to those with untreated OSA [5]. They also reported significant differences in Social Responsiveness Scale scores between children without OSA and those with untreated OSA.
Our observation that social and emotional functioning represented a relative strength, whereas language and communication remained comparatively weaker domains, is consistent with the established neurocognitive profile of children with DS while acknowledging substantial interindividual variability [23].
Improvements in capacities related to daily life activities, such as personal care and domestic activities and to socialization, e.g., interpersonal interactions, play, and cooperative skills, are important to children with DS and their caregivers, furthering their independence and quality of life [28].
Several limitations should be acknowledged. These encouraging results must take into account the small size of the follow-up sample and the observational nature of the study. This follow-up study included only children who were diagnosed with OSA and received treatment either before 36 months (screened group) or after 36 months of age (standard care group), which may limit generalizability.
Moreover, the primary analysis focused on the PP population (NEURO-60M PP), further reducing the number of evaluable participants. The resulting smaller sample size may have introduced selection bias and limited statistical power to detect between-group differences. Nevertheless, the NEURO-60M PP population was considered less susceptible to bias arising from major protocol deviations, including neuropsychological assessments conducted substantially outside the prespecified visit window.
Changes in sleep quality were primarily evaluated using the AHI and OSA severity categories (none, mild, moderate, and severe). However, additional sleep-related parameters that may influence language and communication development were not specifically examined, including primary snoring, total sleep time, sleep latency, wake after sleep onset, and spindle density [1]. Importantly, the observed between-group differences do not appear to be explained by persistent OSA between 36 and 60 months of age.
OSA prevalence and severity were broadly comparable between the groups at 60 months of age, with > 90% of children exhibiting either no OSA or mild OSA in both the groups. Likewise, the longer total sleep time observed in the screened group at 36 months of age was no longer evident at 60 months of age. However, the screened group continued to demonstrate a higher proportion of REM sleep than the standard care group, suggesting that sustained differences in sleep quality may contribute to the prolonged neurocognitive effects observed (Table 3).
Nonetheless, further studies are required to confirm this hypothesis.
In summary, although no significant benefits were observed in language and communication measures in this follow-up study, the early detection and treatment of OSA in infants with DS were associated with long-term improvements in neurocognitive outcomes at 60 months of age. Additional mechanisms and sleep-related factors beyond OSA severity, such as fine sleep microarchitecture, may contribute to verbal development in young children with DS, and these factors can be further investigated in future studies.
Supplementary Information
Below is the link to the electronic supplementary material.ESM 1(DOCX 69.7 KB)